Improving Acetate Tolerance of Escherichia coli by Rewiring Its Global Regulator cAMP Receptor Protein (CRP)

The presence of acetate exceeding 5 g/L is a major concern during E. coli fermentation due to its inhibitory effect on cell growth, thereby limiting high-density cell culture and recombinant protein production. Hence, engineered E. coli strains with enhanced acetate tolerance would be valuable for these bioprocesses. In this work, the acetate tolerance of E. coli was much improved by rewiring its global regulator cAMP receptor protein (CRP), which is reported to regulate 444 genes. Error-prone PCR method was employed to modify crp and the mutagenesis libraries (~3×106) were subjected to M9 minimal medium supplemented with 5–10 g/L sodium acetate for selection. Mutant A2 (D138Y) was isolated and its growth rate in 15 g/L sodium acetate was found to be 0.083 h-1, much higher than that of the control (0.016 h-1). Real-time PCR analysis via OpenArray® system revealed that over 400 CRP-regulated genes were differentially expressed in A2 with or without acetate stress, including those involved in the TCA cycle, phosphotransferase system, etc. Eight genes were chosen for overexpression and the overexpression of uxaB was found to lead to E. coli acetate sensitivity.


Introduction
Acetate, either lignocellulosic-derived or as a fermentative byproduct, can pose a major problem in microbial bioprocesses, especially in the presence of excess glucose [1][2][3][4][5][6]. Being one of the most widely studied growth inhibitors for E. coli fermentations, acetate is known to inhibit cell growth when its concentration exceeds 5 g/L, thereby limiting high cell density and leading to reduced titers or recombinant protein production [7][8][9]. Therefore, it becomes valuable to engineer strains with improved acetate tolerance for sustainable microbial fermentation [10,11]. The protonated form of acetate can penetrate cell membranes, resulting in a reduction of intracellular pH [9] and an accumulation of anions in cell cytoplasm [12], both of which could contribute to growth inhibition.
In order to resolve the undesirable effects of acetate, fermentation conditions have been optimized to reduce acetate formation, such as controlling glucose feed rate, managing dissolved oxygen level, and using fructose instead of glucose as sole carbon source [13]. Acetate toxicity can also be alleviated via the addition of methionine [14], glycine [13], arginine, threonine, and lysine [10].
Reduction in acetate production can also be achieved through the adoption of metabolic engineering tools. Genes related to acetate producing pathways were knocked out [15,16], while heterologous genes have been introduced into E. coli to convert acetate to other less harmful byproducts [17]. In addition, directed evolution of homoserine osuccinyltransferase, an enzyme involved in methionine biosynthesis, was also proved to enhance the acetate tolerance of E. coli [14].
Besides metabolic engineering approaches, classical strain engineering methods of using UV and evolutionary engineering strategies have also been used to improve microbial tolerance towards acetate stress. UV mutagenesis was performed on Clostridium thermoaceticum and Clostridium thermoautophicum [18], whereas acetate-tolerant E. coli and S. cerevisiae mutants were generated through evolutionary engineering [11,19].
Our lab has successfully improved the tolerance of E. coli towards various stresses in the past through engineering its global regulator cAMP receptor protein [32][33][34][35][36]. In this work, we have also chosen cAMP-receptor protein (CRP) as our target regulator to improve the acetate tolerance of E. coli. To our knowledge, this is the first project on engineering a native global regulator to improve strain acetate tolerance. CRP is able to regulate 444 genes in E. coli, especially those involved in carbon metabolism [37,38]. The production and assimilation of acetate is largely controlled by the reactions in the tricarboxylic acid (TCA) cycle, where the genes are mainly regulated by CRP [39,40], which makes CRP an attractive target. Here, we have introduced random mutations to CRP by error-prone PCR. The mutant with the greatest enhancement in acetate tolerance was further characterized in terms of its cross-tolerance to other fermentative byproducts, extracellular acetate concentration and cell morphology studies. The expression changes of 444 CRP-regulated genes were also assessed by quantitative real-time reverse transcription PCR (RT-PCR) using OpenArray ® system.

Library construction
Error-prone PCR was carried out using the GeneMorph® II Random Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA) with the following primers: 5'-GAGAGGATCCATAACAGAGGATAACCGCGCATG-3', and 5'-AGATGGTACCAAACAAAATGGCGCGCTACCAGGTAACGC GCCA-3' (the underlined sequences correspond to BamHI and KpnI restrictions sites respectively), with 30 ng template to yield 1-3 amino acid substitutions per crp gene. The PCR conditions were set as follows: 2 min at 95°C, 30 cycles of 95°C for 1 min, 62°C for 1 min, followed by 72°C for 1 min, and 10 min at 72°C. The resulting crp products were digested and cloned into a digested low copy number (copy number 1-5) self-constructed plasmid pKCP ( Figure S1 & Table S1) and transformed into E. coli DH5α ∆crp cells by electroporation using an Eppendorf® multiporator (Hamburg, Germany).

Mutant selection
The transformants were first cultured in 50 ml M9 minimal medium supplemented with 5 g/L sodium acetate for 20-30 h after inoculation, followed by another two subcultures containing 10 g/L sodium acetate to select for the mutants with improved acetate tolerance. Each subculture was cultivated for 48 h. The cells were plated onto LB agar plates after each round of selection, and individual clones were randomly picked for sequencing. In order to exclude false positives and spontaneous mutations generated during the selection, the mutated crp fragments of the selected plasmids were digested, re-ligated with freshly digested plasmid pKCP and retransformed into fresh E. coli DH5α ∆crp background.

Mutant growth under acetate stress
The growth profile of the CRP mutant isolated from the errorprone PCR libraries was determined at 0, 10 and 15 g/L sodium acetate concentrations. One percent (v/v) overnight cell culture was inoculated into 10-ml M9 medium supplemented with sodium acetate, and cultured at 37°C with orbital shaking at 200 rpm. Their absorbance at 600 nm was recorded at certain time intervals.

Extracellular acetate concentration
One percent (v/v) overnight cell culture was inoculated either into M9 medium (0 g/L sodium acetate) or M9 medium with 10 g/L sodium acetate. To measure the extracellular acetate concentration, 200-µl cell samples were centrifuged at 13,000 rpm for 5 min, and their supernatant was collected. The acetate concentration was determined enzymatically using a Rbiopharm acetic acid test kit (Damstadt, Germany) according to the manufacturer's instructions.

Cross-tolerance to other fermentative products
One percent (v/v) overnight cell culture was inoculated into 10-ml M9 medium supplemented with either 5 g/L sodium formate or 18 g/L sodium propionate. Cells were incubated at 37°C with orbital shaking at 200 rpm, and their OD600 values were measured.

RNA isolation
One percent (v/v) overnight cell culture was inoculated into M9 medium supplemented with or without 10 g/L sodium acetate. Cells were cultured in the absence of sodium acetate for 8 h, or in 10 g/L sodium acetate for 18 h. Total RNA was isolated using the PureLink® RNA mini kit (Life Technologies, Carlsbad, CA, USA), and treated with the PureLink® DNase (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instructions. The quality and integrity of the isolated RNA was determined through spectrophotometer and gel electrophoresis. 800 ng total RNA was converted into cDNA by reverse transcription in a 20-µl reaction mixture using the High Capacity cDNA Reverse Transcription Kit (Life Technologies, Carlsbad, CA, USA) with random primer mix following the manufacturer's protocol.

RT-PCR using the OpenArray® System
RT-PCR for 444 CRP-regulated genes was conducted with the OpenArray® Real-time PCR instrument (Life Technologies, Carlsbard, CA, USA) in duplicates, using the Lightcycler® (Roche, Germany) FastStart® (Roche, Germany) DNA Master SYBR® (Life Technologies, Carlsbard, CA, USA) Green I kit (Roche, Germany) in the reformatted OpenArray® real-time PCR plates (Life Technologies, Carlsbard, CA, USA). The OpenArray® system is a high throughput real-time PCR platform, which allows 224 genes to be assessed on a single plate. 33-nl reaction mixture was loaded into each through-hole on the plates with the OpenArray® AccuFill TM system (Life Technologies, Carlsbard, CA, USA). The default real-time PCR conditions in the provided protocol were followed. The bacterial 16S rRNA (rrsG) was used as the internal standard and the sequence of the primers are given in Table S2. The values of cycle threshold (C t ) were provided by the OpenArray® Realtime qPCR Analysis Software Version 1.0.4. The 2 -ΔΔCt method was utilized to compute the relative expression level. The pvalue for each gene was calculated by student's t-test using the IBM SPSS Statistics Software Version 19.

RT-PCR for genes involved in acetate production
The expression level for genes involved in acetate production (ackA, pta and poxB) was measured in the StepOnePlus TM real-time PCR system (Life Technologies, Carlsbard, CA, USA) using the Power SYBR® Green PCR Master Mix (Life Technologies, Carlsbad, CA, USA). The realtime PCR conditions were carried out as follows: 10 min at 95°C, 40 cycles of 95°C for 15 s, followed by 60°C for 1 min. The bacterial 16S rRNA (rrsG) was used as an internal standard and the sequences of the primers are provided in Table S2. The values of cycle threshold (C t ) were provided by the software, and 2 -ΔΔCt method of relative quantification was utilized to compute the relative expression level. The p-value for each gene was calculated by student's t-test.

Effects of gene overexpression on acetate tolerance
The ASKA clones (GFP-) was used for gene overexpression [42] in E. coli DH5α. One percent (v/v) overnight culture was inoculated into 10-ml M9 medium containing either 0 g/L or 5 g/L sodium acetate. Chloramphenicol (25 µg/ml) was added to maintain the pCA24N-based plasmids, and 0.1 mM IPTG was used to induce the overexpression of genes of interest according to previously established protocol [29]. Cells were incubated at 37°C with orbital shaking at 200 rpm, and their OD600 values were recorded.

Mutant isolation
The mutated crp products obtained via error-prone PCR were cloned into plasmid pKCP. E. coli DH5α ∆crp strain harboring pKCP (containing native crp operon), which was used as the control in this study. Six error-prone PCR libraries with a total library size of ~3×10 6 were subjected to 5-10 g/L sodium acetate for variant selection. The mutant selection process was greatly shortened to 5-7 days as compared to classical strain engineering methods of using adaptive evolution whereby 1,750 generations (~18 days) were required to isolate an acetate-tolerant mutant [19]. Thirty-two colonies were randomly picked and sequenced to determine their amino acid modifications in CRP. Among these 32 clones, eight of them contained mutation Y63F (A1), eight had D138Y (A2), ten contained V176I (A3), and the remaining six clones had modifications at Y99C, A151V, V176A, and L195M (A4). A2 was identified with the best improvement in acetate tolerance out of these four variants and was investigated further in this study.
The amino acid substitution at position D138 of A2 (D138Y) is of particular interest as some other CRP mutants with either improved osmotolerance or 1-butanol tolerance also had this D138 modification [32,35], but unlike A2, those mutants also carried modifications at other locations. D138 is known to play essential roles during the hinge reorientation and helical adjustment upon allosteric activation by cAMP [43] through the formation of hydrogen bonds with G141 [44]. Although D138 is not located in the DNA binding domain (residue 140-209), any modifications to this position may affect the DNA binding affinity of CRP [45].

Mutant growth under acetate stress
The growth profiles of A2 and the control were determined in the absence or presence of acetate stress. When cells were grown in M9 medium without sodium acetate supplement ( Figure 1A), it is found that A2 (0.19 ± 0.002 h -1 ) already outgrew the control (0.13 ± 0.002 h -1 ). The presence of 10 g/L ( Figure 1B) and 15 g/L sodium acetate ( Figure 1C) led to even stronger growth inhibition of the control. Although the growth rate of A2 was significantly higher than the control under acetate stress, the presence of acetate resulted in A2 exhibiting long period of lag phase up to 20 h. With 10 g/L sodium acetate present, the growth rate of A2 was 0.10 ± 0.0004 h -1 , higher than that of the control at 0.026 ± 0.001 h -1 . When sodium acetate concentration was further increased to 15 g/L, the growth rate of A2 became 0.083 ± 0.0003 h -1 , whereas that of the control dropped to 0.016 ± 0.001 h -1 .

Extracellular acetate concentration
The amount of acetate produced by A2 and the control were monitored during cultivation. Without sodium acetate supplement, it was observed that the acetate produced by A2 was higher than the control (Figure 2A). Since the cell growth of A2 was also higher than the control during cultivation, the measured acetate concentration was normalized by OD600 values. The normalized acetate concentration of A2 (average 0.17 g/L per OD600) almost doubled that of the control (average 0.10 g/L per OD600) over the sampling period of 14 h. On the contrary, when A2 and the control were exposed to 10 g/L sodium acetate (~7 g/L acetate), the extracellular acetate concentration was nearly the same ( Figure 2B).
It was reported before that strains with less acetate production would grow faster [46], thus many strategies have been focusing on reducing the production of acetate to improve cell growth under acetate stress [47]. Interestingly, we found that although A2 was able to grow faster than the control in minimal medium, A2 produced more acetate than the control. There are two acetate producing pathways in E. coli: the first is from acetyl-CoA to acetate catalyzed by acetate kinase (ackA) and phosphotransacetylase (pta) with acetyl-phosphate as intermediate, and the second is to convert pyruvate to acetate by pyruvate kinase (poxB) [48]. As demonstrated by our RT-PCR results on ackA, pta and poxB ( Figure S2), A2 had higher expression of all three genes than the control. Therefore, the boosted acetate production in A2 was likely to be attributed to the higher expression of these three genes.

Cross-tolerance to other fermentative byproducts
In addition to acetate, other metabolic byproducts, such as formate and propionate, are present during E. coli fermentation [8,49]. Hence, the tolerance of A2 towards these two byproducts was also investigated. The growth pattern of A2 and the control cultivated in M9 medium supplemented with 5 g/L sodium formate or 18 g/L sodium propionate was presented in Figure 3A and 3B respectively. The growth rate of A2 in formate was 0.16 ± 0.002 h -1 , higher than the control (0.14 ± 0.008 h -1 ), and its performance was much enhanced in the presence of propionate-the growth rate of A2 was 0.07 ± 0.0004 h -1 , doubling that of the control at 0.03 ± 0.0003 h -1 .

Quantitative real-time reverse transcription PCR with an OpenArray® system
We have investigated on the relative expression level of 444 genes regulated by CRP, about 10% of genes in E. coli [37], between mutant A2 and the control in the presence or absence of sodium acetate stress. This was performed to probe the difference of the transcript level of CRP-regulated genes between A2 and the control. Without stress, 407 genes were differentially expressed in A2, including 290 up-regulated genes and 14 down-regulated genes by more than 2-fold (p < 0.05) ( Table S3). The two highest upregulated genes were yfiD and pflB, by 17.7-and 15.5-fold respectively, whereas galE displayed the leading downregulation of 143-fold. On the other hand, the presence of sodium acetate stress resulted in the differential expression of 432 CRP-regulated genes in A2, with 6 upregulated and 406 genes downregulated by more than 2fold at a p-value threshold < 0.05 (Table S4). To be more specific, galE had the biggest increase in A2 by nearly 14-fold, whereas araG, encoding the ATP binding unit of arabinose transporter [37], exhibited the greatest decrease of 27-fold.
It is apparent that the presence of sodium acetate resulted in the downregulation of various functional groups in A2, such as the TCA cycle (sdhCDAB, sucABCD, and mdh), phosphotransferase system (PTS) (ptsG, ptsHI-crr, nagE, and malX), ATP binding cassette (ABC) transporters (malEFG, mglBAC, flhCD, and lsrB), two-component system (ompF, dctA), amino acid metabolism (tnaA, gadA), and amino sugar and nucleotide sugar metabolism (nanE) ( Table 1). Many of these genes have already been demonstrated to be associated with cell adaptation to acetate stress [39,49,50]. Interestingly, except for the aforementioned genes involved in the TCA cycle, the rest of the genes were all upregulated when sodium acetate was absent. The downregulation of TCA cycle in the absence of sodium acetate may lead to high acetate accumulation in A2 [47].
Acetate can also be utilised as carbon source, whereby the uptake of acetate is performed by acetyl-CoA synthetase (acs), and metabolized through the glyoxylate cycle (aceBAK operon, mdh, gltA and acnB). Many studies on the response of E. coli towards acetate stress have reported that acs and aceBAK are   induced by acetate [46]. However, as presented in Table 1, we found that the expression of acs was downregulated by over 7fold, and the genes from glyoxylate pathway had repressed expression in A2 with acetate stress present. Acetate can lead to internal acidification whereby the dissociation of weak acids in cytoplasm causes decline in intracellular pH [12,51]. Genes such as yfiD, pflB, gadA and gadB are able to increase their expression to protect E. coli against internal acidification (Table 1) [50,52], which concurs with our findings in this study. In particular, yfiD, which encodes stress-induced alternate pyruvate formate lyase subunit, was upregulated by almost 17-fold even without the addition of sodium acetate. The expression of pflB (pyruvate formate lyase, 15.5-fold), gadA (2.6-fold) and gadB (glutamate decarboxylase, 2.3-fold) also enhanced in A2, suggesting that A2 was protected against internal acidification by the preprogrammed upregulation of these genes. The upregulation of yfiD, pflB and gadA (approximately 2-to 3-fold) in the presence of sodium acetate might ensure the continual protection of A2 against internal acidification caused by acetate anion. Yet, we found that the overexpression of individual genes such as pflB, yfiD, or gadA did not confer acetate tolerance to E. coli (data not shown), indicating that coordinated changes from multiple genes might be required [53]. Previous DNA microarray analysis of E. coli after long-term adaption to acetate stress illustrated that certain genes involved in the uptake and catabolism of carbon and energy sources, except for glucose, decreased in their mRNA abundances, such as maltose (malXY, and malEFG), galactose (galETKM, galS, and mglBAC), amino sugars (nanATEK-yhcH), sugar alcohols (srlA1A2BD-gutM, and srlR-gutQ), amino acids (tnaA), and dicarboxylic acids (dctA) to adapt to acetate stress [49]. Our results on A2 also display repression of these genes under acetate stress. Another similar finding is that the expression of transport genes such as malE and ompF are both repressed in the presence of acetate [8,50].

Effects of gene overexpression on acetate tolerance
Eight genes with the largest fold-change in their expression level in A2 under acetate stress (Table S4), including the upregulated genes pflB, yfiD, galE, gadA (2-to 14.3-fold), and the downregulated genes uxaB, feaR and araG (-22-to -27.3fold), were chosen for overexpression. E. coli DH5α harboring pCA24N plasmid was adopted as control. It was observed that the overexpression of pflB, yfiD, galE and gadA did not improve the growth of E. coli DH5α when facing stress, and the overexpression of feaR and araG did not result in acetate sensitivity (data not shown). In contrast, a noticeable growth difference appeared between the control and uxaB overexpression in the presence of acetate stress. Although both had similar growth rate of 0.09-0.1 h -1 when cultured without sodium acetate ( Figure 4A), the presence of 5g/L sodium acetate reduced the growth of the control to ~ 0.05 h -1 and exerted a greater deleterious effect on the cells overexpressing uxaB ( Figure 4B)-its cell growth rate was reduced by ~60% to 0.021 h -1 . uxaB (altronate oxidoreductase) is responsible for the reversible NADH-dependent reduction of D-tagaturonate to D-altronate in galacturonate catabolism pathway [54]. To our knowledge, there have been no reports to date relating uxaB with acetate metabolism or tolerance, which may require further investigation in the future.

Conclusions
In this study, we have successfully isolated an E. coli mutant strain with much enhanced tolerance towards acetate stress by rewiring its global regulator CRP. Error-prone PCR was adopted to introduce mutations to CRP and the mutant selection process was greatly shortened to several days as compared to classical strain engineering methods of using adaptive evolution. To the extent of our knowledge, this is the first study whereby a native regulator is engineered by random mutagenesis method for improved cell performance under acetate stress.  (DOC) Table S3. CRP-regulated genes with >2-fold change in their expression level in A2 as compared to the control in the absence of sodium acetate stress, using a p-value threshold less than 0.05.
(DOC) Table S4. CRP-regulated genes with >2-fold change in their expression level in A2 as compared to the control in the presence of sodium acetate stress, using a p-value threshold less than 0.05. (DOC)